Mutation of Position 52 in ERK2 Creates a Nonproductive Binding Mode for Adenosine 5′-Triphosphate†,‡ Megan J

+ +

Biochemistry 1996, 35, 5641-5646 5641

Mutation of Position 52 in ERK2 Creates a Nonproductive Binding Mode for Adenosine 5′-Triphosphate†,‡ Megan J. Robinson,§ Paul C. Harkins,| Jiandong Zhang,| Richard Baer,⊥ John W. Haycock,# Melanie H. Cobb,§ and Elizabeth J. Goldsmith*,| Departments of Pharmacology, Biochemistry, and Microbiology, UniVersity of Texas Southwestern Medical Center, Dallas, Texas 75235, and Department of Biochemistry, Louisiana State UniVersity, New Orleans, Louisiana 70119 ReceiVed NoVember 15, 1995; ReVised Manuscript ReceiVed February 16, 1996X

ABSTRACT: Among the protein kinases, an absolutely conserved lysine in subdomain II is required for high catalytic activity. This lysine is known to interact with the substrate ATP, but otherwise its role is not well understood. We have used biochemical and structural methods to investigate the function of this lysine (K52) in phosphoryl transfer reactions catalyzed by the MAP kinase ERK2. The kinetic properties of activated wild-type ERK2 and K52 mutants were examined using the oncoprotein TAL2, myelin basic protein, and a designed synthetic peptide as substrates. The catalytic activities of K52R and K52A ERK2 were lower than that of wild-type ERK2, primarily as a consequence of reductions in kcat. Further, there was little difference in Km for ATP, but the Km,app for peptide substrate was higher for the K52 mutants. The three-dimensional structure of unphosphorylated K52R ERK2 in the absence and presence of bound ATP was determined and compared with the structure of unphosphorylated wild-type ERK2. ATP adopted a well-defined but distinct binding mode in K52R ERK2 compared to the binding mode in the wild-type enzyme. The structural and kinetic data show that mutation of K52 created a nonproductive binding mode for ATP and suggest that K52 is essential for orienting ATP for catalysis.

A lysine in subdomain II is absolutely conserved among enzymes (Snyder et al., 1985; Kamps & Sefton, 1986; Gibbs the protein kinase family members (Hanks et al., 1988). This & Zoller, 1991). In the case of the MAP kinase ERK2, lysine is required for efficient phosphoryl transfer reactions however, mutation of the subdomain II lysine, K52, to (Kamps & Sefton, 1986; Gibbs & Zoller, 1991). An arginine creates a dominant interfering kinase in intact cells examination of the structures of ATP-bound cyclic AMP- (Sontag et al., 1993; Frost et al., 1994), although the mutant dependent protein kinase (cAPK),1 cyclin-dependent kinase retains approximately 5% of wild-type activity (Posada & 2 (CDK2), and extracellular signal-regulated protein kinase Cooper, 1992; Robbins et al., 1993). Surprisingly, introduc- 2 (ERK2) indicates that this lysine in subdomain II is tion of the same mutation in the closely related kinase ERK1 positioned neither to act as the catalytic base nor to stabilize yields a protein without detectable activity (Robbins et al., the γ-phosphoryl group in the transition state. Nevertheless, 1993). The low but measurable kinase activity of K52R covalent modification and mutational analysis show that this ERK2 makes ERK2 a good subject for examining the impact lysine residue is essential for catalytic activity (Zoller & of this lysine on enzyme kinetics; in contrast to “kinase- Taylor, 1979; Zoller et al., 1981; Kamps & Sefton, 1986; dead” mutants of other protein kinases, quantitative informa- Buechler & Taylor, 1989; Gibbs & Zoller, 1991). tion can be obtained with K52R ERK2. To examine the Mutation of the conserved lysine in several protein kinases, generality of data obtained with K52R ERK2, K52 was also notably cAPK and v-Src, resulted in residual protein kinase mutated to alanine and the properties of this mutant were activity estimated to be less than 0.1% of the wild-type characterized. We report here biochemical data on K52R and K52A ERK2 and structural analysis of K52R ERK2 † This work was supported by a postdoctoral fellowship from the complexed with ATP. Arthritis Foundation (M.J.R.), grants to M.H.C. (I1243) and E.J.G. (I1228) from the Welch Foundation, and grants to M.H.C. (DK34128), E.J.G. (DK46893), and J.W.H. (NS25134 and MH00967) from the MATERIALS AND METHODS National Institutes of Health. ‡ Coordinates have been deposited in the Brookhaven Protein Data Purification of Wild-Type, K52R, and K52A ERK2. K52 Bank (filename 1GOL). was mutated to alanine using the Stratagene Chameleon * To whom correspondence should be addressed: Phone (214) 648- 5009; Fax (214) 648-6336. double-stranded mutagenesis kit. The presence of the § Department of Pharmacology, University of Texas Southwestern mutation was confirmed by DNA sequencing. K52R ERK2 Medical Center. was made previously (Robbins et al., 1993). Histidine- | Department of Biochemistry, University of Texas Southwestern Medical Center. tagged ERK2 proteins were expressed and purified from ⊥ Department of Microbiology, University of Texas Southwestern Escherichia coli as previously described (Robbins et al., Medical Center. 1993). All proteins were stored at -80 °C. # Louisiana State University. X Abstract published in AdVance ACS Abstracts, April 1, 1996. Purification of ConstitutiVely ActiVe MEK1 (MEK1R4F). 1 Abbreviations: ERK2, extracellular signal-regulated protein kinase Constitutively active MEK1 in the pRSET expression vector 2; cAPK, cyclic AMP-dependent protein kinase; CDK2, cyclin- was the generous gift of Dr. Natalie Ahn (Mansour et al., dependent protein kinase 2; MEK, MAP/ERK kinase; MBP, myelin 1993). It was expressed in the BL21-DE3 strain of E. coli basic protein; GST, glutathione S-transferase; MEK1R4F, a constitutively active form of MEK1; ERKtide, peptide substrate for ERK1/2 from that vector or a modified pET16b vector encoding with sequence ATGPLSPGPFGRR. additional histidine residues to improve binding of the S0006-2960(95)02723-1 CCC: $12.00 © 1996 American Chemical Society

+ + + +

5642 Biochemistry, Vol. 35, No. 18, 1996 Robinson et al. histidine-tagged protein to the nickel resin. The bacteria were concentration of 0.5 mg/mL or ERKtide at 2 mM. The grown at room temperature in terrific broth in the presence concentration of ATP was determined by optical density. of 0.02 mg/mL chloramphenicol and 0.1 mg/mL ampicillin Reactions were incubated for 30 min (MBP), 1 h (TAL2), until the OD600 was approximately 0.4-0.6. Cells were or 10 min (ERKtide) as 30 °C and stopped by the addition treated with isopropyl â-D-thiogalactopyranoside (0.05 mM) of concentrated sample buffer. Reactions with MBP or overnight at room temperature prior to harvest by centrifuga- TAL2 were analyzed by electrophoresis, and bands corre- tion, washed in sonication buffer (50 mM sodium phosphate, sponding to MBP or TAL2 were excised and counted by pH 8.0, 300 mM NaCl), and pelleted. The pellets were liquid scintillation. frozen in liquid nitrogen and thawed on ice. Thirty-five Assays containing ERKtide were analyzed using P81 milliliters of sonication buffer with protease inhibitors was paper. ERK2 and mutants were activated without [γ-32P]- added per liter of original bacterial culture. Cells were lysed ATP to eliminate counts on the P81 paper due to ERK2. using a Stanstead cell disrupter, and the lysate was clarified Cpm bound to P81 paper in assays from which ERKtide was at 30000 rpm for 30 min in a Beckman Ti35 rotor. The omitted were subtracted as background. Reactions were supernatant was adjusted to 1 M NaCl, 1% Tween 20, and stopped by spotting on P81 paper. Papers were washed in 20% glycerol, then mixed with 0.5 mL of nickel NTA- dilute phosphoric acid, dried, and counted by liquid scintil- agarose resin (Qiagen), and rocked at 4 °C for 2 h. The lation (Witt & Roskoski, 1980; Roskoski, 1983). resin was washed with 10 mL of 50 mM sodium phosphate, Apparent Km and Vmax values were derived by fitting the pH 8.0, 10% glycerol, and 0.25% Tween 20 and then with data into the Michaelis-Menten equation with the aid of 10 mL of 20 mM imidazole in sonication buffer. MEK1R4F Enzfitter (Elsevier-Biosoft) and kcat (Biometallics, Inc.) was eluted with a 40 mL gradient of 20-250 mM imidazole, kinetic analysis software. pH 7.0, in sonication buffer. Fractions with activity and Determination of the Crystal Structure of K52R ERK2. immunoreactivity were pooled and dialyzed overnight against Crystals of K52R ERK2 were grown as previously described purification buffer (20 mM Tris, pH 7.5, 1 mM DTT, 1 mM (Zhang et al., 1993) by the vapor diffusion method, EGTA, 10 mM benzamidine, and 10% glycerol). The equilibrating a 6 mg/mL solution of the protein against 18% protein was then applied to a MonoQ HR 5/5 column PEG 8000 and 0.2 M (NH4)2SO4. The complex with ATP (Pharmacia) equilibrated in purification buffer and eluted was prepared by soaking protein crystals in a stabilizing with a 0-0.4 M gradient of NaCl in purification buffer. No solution that contained 2 mM MgCl2, 10 mM ATP, and 50 impurities were detected in the constitutively active, mM Tris, pH 7.0, for 4 h. The crystals of the mutant protein MEK1R4F preparation by staining with Coomassie blue. were isomorphous with those of wild-type ERK2. Wild-type MEK1 phosphorylated with MEKK1 as described Data were collected to 2 Å at 20 °C on an RAXIS (Rigaku) (Xu et al., 1995) was generously provided by Andrei image plate detector using mirror monochromated Cu KR Khokhlatchev. X-rays generated by a Rigaku RU200 rotating anode (50 kV, ActiVation of ERK2 Proteins with MEK1R4F or Phos- 100 mA). The collected data were reduced using DENZ0 phorylated MEK1. Phosphorylation reactions contained and SCALEPACK (Minor, 1993; Otwinoskwi, 1993). All 20 mM Tris, pH 8.0, 1 mM DTT, 1 mM benzamidine, 10 crystallographic refinement was done in X-PLOR (Bru¨nger, mM MgCl , 100 µM ATP, 2-30 µg/mL ERK2 protein, and 2 1995), and model building was done using XtalView 0.2 unit of MEK1R4F or MEK1 and were incubated for (McRee, 1994), with the exception of the final structure. As 1.5 h at 30 °C (Mansour et al., 1994; Xu et al., 1995). a test of the quality of the final structure, a simulated After this time there was no further activation of ERK2. annealing omit map (SA-omit; Hodel et al., 1992) was Reactions were then placed on ice until use in substrate calculated in X-PLOR; the mutant side chain and Mg‚ATP assays. The stoichiometries of phosphorylation, 0.6-0.8 were removed from the calculation. The map and model mol of phosphate/mol of ERK2, were equivalent∼ for ERK2 were then visualized in “O” (Jones et al., 1991) (see Figure and mutants. 2A). Kinetic Analysis of ERK2 and Mutants. A synthetic peptide (ATGPLSPGPFGRR, ERKtide) was designed on the RESULTS basis of the ERK1 substrate sequence preference from a peptide library screen (Z. Songyang and L. C. Cantley, Kinetic Analysis of Wild-Type, K52R, and K52A ERK2. submitted) and residues surrounding the ERK phosphoryla- ERK2 phosphorylates a variety of substrates including MBP, tion site in tyrosine hydroxylase (Haycock et al., 1992). tyrosine hydroxylase, c-Elk, c-Myc, c-Jun, and phospholipase ERKtide was dissolved in water. Human GST-TAL2 (53- A2 (Erickson et al., 1990; Alvarez et al., 1991; Gille et al., 108) (Xia et al., 1994) was purified using the method of 1992; Haycock et al., 1992; Lin et al., 1993; Nemenoff et Smith and Johnson (1988). Myelin basic protein (MBP) al., 1993). It was necessary to identify substrates that (0.6-30 µM) (Sigma), TAL2 (0.625-25 µM), or ERKtide contained a single phosphorylation site and were available (0.1-5 mM) was added to reactions containing 20 mM Tris, in quantity. MBP was phosphorylated not only on T97, the pH 8.0, 1 mM DTT, 1 mM benzamidine, 10 mM MgCl2, published site (Erickson et al., 1990), but also on T94 at 1 200 µM ATP ([γ-32P]ATP, 25 µCi/reaction), and 0.2-3.0 mM ATP (data not shown); thus it was useful only for µg/mL ERK2 protein that was freshly activated by phos- qualitative comparisons. The TAL1 and TAL2 oncoproteins phorylation with MEK1R4F or phosphorylated MEK1 as were tested because they each contain a single ERK described above. No phosphorylation of substrate occurred phosphorylation site (Cheng et al., 1993; Xia et al., 1994). in the presence of MEK1 or MEK1R4F alone (data not Of several TAL1 and TAL2 fragments tested, a GST-fusion shown). To determine kinetic parameters for ATP, assays protein, containing the 56 carboxy-terminal residues of were performed as above except with 1 µM to 1 mM ATP human TAL2, was the best ERK2 substrate, although it was ([γ-32P]ATP, 20 µCi/reaction) andeither MBP or TAL2 at a phosphorylated at only about 5% of the rate of MBP (Table

+ + + +

Nonproductive ATP Binding Mode of K52R ERK2 Biochemistry, Vol. 35, No. 18, 1996 5643

Table 1: Kinetic Constants of Wild-Type and Mutant ERK2sa a peptide library screen (Z. Songyang and L. C. Cantley, submitted) and with certain residue choices based on the wild-type K52R K52A ERK2 ERK2 ERK2 substrate tyrosine hydroxylase (Haycock et al., 1992). TAL2b Previous experiments with MBP as substrate indicated that c Km,app[ATP] (µM) 220 ( 60 550 ( 130 N/D the specific activity for wild-type ERK2 was 20-fold -1 c ∼ kcat,app (min ) 1.4 ( 0.15 >0.009 ( 0.001 N/D greater than that for the K52R mutant (Robbins et al., 1993). c c Km,app[TAL2] (µM) >40 N/D N/D -1 d e e c To examine this difference more thoroughly, we determined kcat,app/Km,app (min /µM) 0.31 0.0016 N/D ERKtidef apparent Vmax and Km values for ATP and a peptide substrate Km,app[ATP] (µM) 350 ( 60 400 ( 80 380 ( 160 for the phosphorylated forms of wild-type ERK2, K52R -1 kcat,app (min ) 120 ( 8 >6.0 >1.0 ERK2, and K52A ERK2 (Table 1). With both GST TAL2 g h - Km[peptide] (µM) 450 ( 230 >5000 >5000 -1 d e c and ERKtide, the Km,app[ATP] of K52R ERK2 was not kcat,app/Km,app (min /µM) 0.27 0.0004 N/D MBPi significantly higher than that of wild-type ERK2. With Km,app[ATP] (µM) 190 ( 97 270 ( 30 300 ( 70 ERKtide, the Km,app[ATP] of K52A ERK2 was also indis- -1 kcat,app (min ) 80 ( 19 0.38 ( 0.02 0.88 ( 0.1 tinguishable from wild-type ERK2. On the other hand, the Km[MBP] (µM) 55 ( 25 4 ( 1.5 1.8 ( 0.6 -1 k with each substrate was much lower for K52R ERK2 kcat,app/Km,app (min /µM) 1.47 0.09 0.49 cat a and K52A ERK2 than for wild-type ERK2. The kcat for Kms for ATP were measured at 30 µM MBP, 25 µM TAL2, ∼ ∼ K52R ERK2 was 0.5-5% and for K52A ERK2 1% that and 2 mM ERKtide. Kms for phosphoryl acceptors were measured at ∼ ∼ 200 µM ATP. b TAL2 data are the result of five to six independent of the wild-type enzyme for each peptide/protein substrate. experiments. Only at the higher TAL2 substrate concentrations tested However, the ratio of kcat/Km,app for wild-type, K52R, and could Km,app data be quantitated with the K52R mutant, and very little K52A ERK2 varied depending on the substrate (see Table activity was detected even at the highest concentration of TAL2 tested 1). with K52A ERK2 (25 µM) (Table 1). In addition, it was not possible to test higher TAL2 concentrations, as the protein solution could not On further analysis, the variability in peptide kcat/Km,app be concentrated any further. c N/D ) not enough detectable activity or was accounted for at least in part by the finding that the d data points available for quantitation. kcat/Km values represent phos- apparent K values for the peptide substrate are higher for phoryl acceptor, not ATP. e These values were determined by calculat- m K52R ERK2 and K52A ERK2 than for wild-type ERK2. ing the slopes of the lines plotted from the V0 vs [s] over the entire concentration range tested. f ERKtide data are the result of two to four The Km,app[ERKtide] of wild-type ERK2 was 0.5 mM. In g ∼ independent experiments. Estimates were based on the fact that, at comparison, the Km,app[ERKtide] values of K52R and K52A the highest concentration tested, 5 mM, only slight curve saturation ERK2 are greater than 5 mM, based on the observation that h was evident. Actual Km is likely to be at least 2-fold higher. The Km the rates were not approaching saturation at 5 mM, the must be higher than 5 mM as K52A is only beginning to show activity at this concentration. The actual value may be at least 5-fold higher. highest peptide concentration that could be tested. As MBP i MBP data are the result of two to five independent experiments. is not phosphorylated on a single site, the Km values for MBP of the K52 mutants and wild-type ERK2 cannot be directly compared. Thus, the kinetic findings suggest that mutagenesis of K52 affects both kcat and Km for peptide substrates.

Structural Analysis of K52R ERK2.A FK52R - FK52R-ATP difference map using phases from wild-type ERK2 (Zhang et al., 1994) clearly shows density for an ATP molecule, as well as density suggesting that R52 interacts with bound ATP. A model for the ATP plus a Mg2+ was fit into the density and subjected to positional refinement in X-PLOR (see the legend for Figure 2). Figure 2A shows a difference map calculated with phases derived from the final model, omitting Mg‚ATP and the R52 side chain from refinement and phase calculations. As in wild-type ERK2, ATP binds between the two domains of the enzyme (Figure 2B). The ribose takes up a 3′-endo, 2′-exo conformation. Many of the ATP binding interactions are identical to those observed in the wild-type enzyme; the adenine is in van der Walls contact with L154,

FIGURE 1: Phosphorylation of MBP (squares) and TAL (circles) I82, I29, V37, and M106. The adenine N1 is hydrogen- by activated wild-type ERK2 (closed) or by activated K52R ERK2 bonded to the M106, and N6 is hydrogen-bonded to D104. (open). The plot shows picomoles of ATP incorporated into the The ribose is oriented by hydrogen bonds of the O2′ and substrate over time. Due to the low amount of phosphorylation of O3′ hydroxyl groups to D109 and K112, respectively, TAL2 by the activated wild type and of both proteins by activated forming a hydrogen-bonding network. K52R ERK2 shown in this figure, only values for the activated enzymes could be measured for kinetic analyses. Furthermore, in The position of the triphosphate moiety, however, is contrast to data with MBP, phosphorylation of TAL2 reached but different in the mutant enzyme as a result of direct interac- did not exceed a stoichiometry of 0.4 mol of phosphate/mol of tions with the guanidinium group of R52. R52 is positioned TAL2 (data not shown). Phosphorylation of TAL2 by activated K52R ERK2 is reproduced in the inset to better show these data. differently from K52 in the wild-type kinase and is extended into the active site cleft (Figure 2B,C). The arginine side 1, Figure 1). A synthetic peptide (ERKtide) was designed chain is hydrogen-bonded via a water molecule to N7 of the on the basis of the work of Songyang et al. (1995), who adenine ring in its major conformation (see Figure 2B) and determined an optimum substrate sequence for ERK1 using is ion paired with the R-phosphate oxygens; large differences

+ + + +

5644 Biochemistry, Vol. 35, No. 18, 1996 Robinson et al.

FIGURE 2: A model for Mg‚ATP was fit into the density and subjected to positional refinement in X-PLOR. After refinement, the model was verified and rebuilt into a simulated annealing omit map calculated with the ATP, the Mg2+, and the R52 side chain removed (panel A). The blue contours are 1.5σ; the magenta contours are 4σ. Density in the loop region of residues 29-36 suggested a movement of the loop, and the model was rebuilt in this region to fit the density. Subsequent to model verification and rebuilding, a round of simulated annealing (Tmax ) 500 °C), as well as positional and B-factor refinements, was carried out, and the final model was verified by 2Fo - Fc and difference maps. Panel B: Stereoview of the differences between the wild type ERK2 and the K52R ERK2 phosphate binding loop are shown. K52R is in magenta while wild-type ERK2 is drawn in green. The ATP shown is from the K52R refinement. The magnesium ion is shown in yellow. Panel C: Stereoview of the difference in binding mode for the ATP between K52R ERK2 and wild-type ERK2 is illustrated. [Panels B and C rendered using SETOR (Evans, 1993).] are observed in the C5′-O5′ bond angle of ATP bound to a hydrogen bond with the guanidinium group of R65. wild-type (160°) and K52R ERK2 (-90°). The â- and From these differences in structure, it is clear that the γ-phosphate moieties take up different positions as compared introduction of the mutation K52R severely alters the position to the wild-type enzyme, such that the γ-phosphate is in the assumed by the triphosphate of the substrate ATP, destroys position to be hydrogen-bonded to the -amino group of conserved side chain-side chain interactions, and disrupts K149 and the hydroxyl group of S151. Arginine at position the coordination sphere of Mg2+. These structural results 52 does not form the ion pair with E69, which is observed are consistent with the conclusion that this mutation has for K52 in wild-type ERK2 (Figure 2). created a nonproductive binding mode for ATP. Interest- Concomitant with the rearrangement of the triphosphate ingly, no structural differences between wild-type and K52R in the K52R mutant, the Mg2+ is coordinated differently. In ERK2 other than those discussed above were found. It is the mutant, Mg2+ is coordinated by oxygens from all three unlikely, however, that the K52A mutation will introduce of the phosphate groups, the carbonyl oxygen of N152, and the same structural changes in ATP binding because most the carboxylate group of D165. In the K52R mutant the of the observed structural changes in K52R ERK2 arise from phosphate binding ribbon (residues 30-35) is better ordered an overlap of the arginine side chain with the ATP binding than in the wild type, and Y34 at the apex of the turn makes site.

+ + + +

Nonproductive ATP Binding Mode of K52R ERK2 Biochemistry, Vol. 35, No. 18, 1996 5645

DISCUSSION ERK2. Interestingly, however, the Km values for ATP of ERK2 and both K52 mutants are indistinguishable. Because Role of the ConserVed Lysine in Subdomain II. ERK2 of the differences in interactions of ATP with these different and the protein kinases share a surprising number of ERK2 molecules discerned and predicted from the structures, conserved residues, including K52 in subdomain II, D147, perhaps K * K . This is consistent with results of Adams K149, and N152 in subdomain Vlb, and D165 in subdomain m s and Taylor (1992), who reported that the rate-limiting step VII. The functions of D147, which serves as the catalytic in cAPK was release of product ADP. base, and D165, which coordinates Mg2+, are relatively Structurally in K52R ERK2, the ATP is well ordered in certain (Knighton et al., 1991; Cole et al., 1995). The the binding cleft. Nevertheless, the combination of structural functions of other residues, such as the lysine in subdomain changes induced by the arginine side chain, most noticeably II, have remained obscure. Mutation of the conserved lysine the altered triphosphate conformation and concomitant in subdomain II has become a standard method of inactivat- change in Mg2+ coordination, leads to a nonproductive, ing protein kinases. For example, Kamps and Sefton found bound complex in which ATP cannot assume the conforma- that neither histidine nor arginine could replace this lysine tion required for phosphoryl transfer. However, while the to support the protein kinase activity or transforming potential observed structural changes may have been caused by steric of Src (Kamps & Sefton, 1986). Comparable mutants of problems from accommodation of the larger arginine side cyclin-dependent kinases, protein kinase C, MAP/ERK chain in K52R ERK2, the similar kinetic profile of K52A kinase, and other protein kinases have also been shown to ERK2 suggests that substitution with this small uncharged be nonfunctional in ViVo (Booher & Beach, 1986; Jove et side chain also may result in a nonproductive ATP binding al., 1987; Ohno et al., 1990; Lange-Carter et al., 1993; mode. Although the ATP position and conformation in the Mansour et al., 1994). low activity (unphosphorylated) structure is most likely Alanine scanning mutagenesis of yeast cAPK demon- different from that in the high activity (doubly phosphory- strated that mutation of the lysine in subdomain II was one lated form), a similar deviation in the ATP binding mode of of the most effective of the approximately 65 mutations tested the mutants as compared to wild-type ERK2 is expected to in reducing kinase activity (Gibbs & Zoller, 1991). Exami- occur in both the low activity and high activity enzyme nation of the kinetic parameters of crude yeast cell extracts containing the mutated yeast cAPK enzymes showed that structures. mutation of the conserved lysine to alanine resulted in k / We have shown that the mutation of the conserved lysine cat in subdomain II creates a nonproductive binding mode for Km [ATP] that was 0.02% of wild type, the lowest activity observed with any∼ of the cAPK alanine-scanning mutants. ATP, thus drastically lowering the activity of the enzyme. Interestingly, disruption of the ATP binding mode is a The decrease in activity was due mostly to changes in kcat, however, suggesting effects on catalysis and not just ATP mechanism of regulation of protein kinases. In CDK2, it recognition and binding (Taylor et al., 1993). Our data also was found that the low activity conformation of that molecule lead to the conclusion that mutation of this residue has a has a nonproductive binding mode for ATP (DeBondt et al., 1993); the position of the residue equivalent to K52 is also negligible effect on Km for ATP but has a marked effect on affected. Similarly, in the insulin receptor and twitchin kcat, suggesting that interactions between ATP and K52 are important to binding of the transition state structure or kinase the low activity forms have a blocked ATP binding facilitating conformational changes needed to promote the site (Hu et al., 1994; Hubbard et al., 1994). Apparently, it phosphoryl transfer reaction. Our results further demonstrate is easy to induce structural changes that interfere with the unexpected but significant changes in Km,app for peptide. The ATP binding site and thereby impair protein kinase function. mechanism by which mutation of K52 influences peptide The exquisite sensitivity to the position and conformation substrate binding could be a direct one in which the peptide of ATP may serve a regulatory role in preventing protein is blocked by the displaced ATP. Alternatively, K52 may kinases from hydrolyzing ATP in the absence of protein have an indirect role in peptide substrate binding, perhaps substrates. by promoting domain rotation that cannot occur in the We hope to be able to better understand the function of mutant. The similarity of the effects on peptide substrate the several conserved catalytic residues in ERK2 and in all Km caused by mutation of K52 to arginine or alanine is protein kinases when the structure of phosphorylated ERK2 evidence against K52 having a direct role in peptide substrate is solved. binding. It is highly improbable that the lysine side chain either ACKNOWLEDGMENT stabilizes the γ-phosphoryl transition state intermediate or We thank Natalie Ahn and Sam Mansour (University of acts as the base catalyst. In the crystal structure of the ternary Colorado, Boulder) for providing the MEK1R4F construct, complex of cAPK (cAPK + Mg‚ATP + protein kinase Andrei Khokhlatchev for providing active MEK1 protein, inhibitor), it was noted that the residue equivalent to K52 is Pequin Wu for help with mutagenesis and purification of within 3.5 Å of both the R- and â-phosphates, not the K52A ERK2, Meg Phillips and Joe Albanesi for critical γ-phosphoryl. It is possible that the role of the lysine is to reading of the manuscript, and Jo Hicks for help with neutralize charges on ATP and to orient the triphosphate for preparation of the manuscript. We also thank Meg Phillips maximum catalytic efficiency. Contacts made by the argi- for invaluable advice with the kinetic studies. nine side chain account at least in part for the differences in the ATP binding mode of K52R ERK2. These contacts REFERENCES would not be expected with alanine in the K52A mutant Adams, J. A., & Taylor, S. S. (1992) Biochemistry 31, 8516-8522. because of the smaller size of the alanine side chain. The Alvarez, E., Northwood, I. C., Gonzalez, F. A., Latour, D. A., Seth, structural data suggest that there might be differences in A., Abate, C., Curran, T., & Davis, R. J. (1991) J. Biol. Chem. affinity for ATP among ERK2, K52R ERK2, and K52A 266, 15277-15285.

+ + + +

5646 Biochemistry, Vol. 35, No. 18, 1996 Robinson et al.

Booher, R., & Beach, D. (1986) Mol. Cell. Biol. 6, 3523-3530. McRee, D. (1994) XtalView 2.0, San Diego Computer Center, San Bru¨nger, A. (1995) X-PLOR, Version 3.1, Yale University Press, Diego, CA. New Haven, CT. Minor, W. (1993) XDISPLAYF Program, Purdue University, Buechler, J. A., & Taylor, S. S. (1989) Biochemistry 28, 2065- Lafayette, IN. 2070. Nemenoff, R. A., Winitz, S., Quian, N. X., Van Putten, V., Johnson, Cheng, J.-T., Cobb, M. H., & Baer, R. (1993) Mol. Cell. Biol. 13, G. L., & Heasley, L. E. (1993) J. Biol. Chem. 268, 1960-1964. 801-808. Ohno, S., Konno, Y., Akita, Y., Yano, A., & Suzuki, K. (1990) J. Cole, P. A., Grace, M. R., Phillips, R. S., Burn, P., & Walsh, C. T. Biol. Chem. 265, 6296-6300. (1995) J. Biol. Chem. 270, 22105-22108. Otwinowski, Z. (1993) Oscillation Data Reduction Program, in DeBondt, H. L., Rosenblatt, J., Jancarik, J., Jones, H. D., Morgan, Proceedings of the CCP4 Study Weekend: Data Collection and D. O., & Kim, S.-H. (1993) Nature 363, 595-602. Processing (compiled by L. Sawyer, M. Isaacs, & S. Bailey) pp Erickson, A. K., Payne, D. M., Martino, P. A., Rossomando, A. J., 56-62, SERC Daresbury Laboratory, England. Shabanowitz, J., Weber, M. J., Hunt, D. F., & Sturgill, T. W. (1990) J. Biol. Chem. 265, 19728-19735. Posada, J., & Cooper, J. A. (1992) Science 255, 212-215. Evans, S. V. (1993) J. Mol. Graphics 11, 134-138. Robbins, D. J., Zhen, E., Owaki, H., Vanderbilt, C. A., Ebert, D., Frost, J. A., Geppert, T. D., Cobb, M. H., & Feramisco, J. R. (1994) Geppert, T. D., & Cobb, M. H. (1993) J. Biol. Chem. 268, 5097- Proc. Natl. Acad. Sci. U.S.A. 91, 3844-3848. 5106. Gibbs, C. S., & Zoller, M. J. (1991) J. Biol. Chem. 266, 8923- Roskoski, R. (1983) Methods Enzymol. 99, 3-6. 8931. Scott, A., Haystead, C. M. M., & Haystead, T. A. J. (1995) J. Biol. Gille, H., Kortenjann, M., Thomae, O., Moomaw, C., Slaughter, Chem. 270, 24540-24547. C., Cobb, M. H., & Shaw, P. E. (1995) EMBO J. 14, 951-962. Smith, D. B., & Johnson, K. S. (1988) Gene 67, 31-40. Hanks, S. K., Quinn, A. M., & Hunter, T. (1988) Science 241, 42- Snyder, M. A., Bishop, J. M., McGrath, J. P., & Levinson, A. D. 52. (1985) Mol. Cell. Biol. 5, 1772-1779. Haycock, J. W., Ahn, N. G., Cobb, M. H., & Krebs, E. G. (1992) Songyang, Z., Blechner, S., Hoagland, N., Hoekstra, M. F., Piwnica- Proc. Natl. Acad. Sci. U.S.A. 89, 2365-2369. Worms, H., & Cantley, L. C. (1995) Curr. Biol. 4, 973-982. Hodel, A., Kim, S.-H., & Bru¨nger, A. (1992) Acta Crystallogr. Sontag, E., Federov, S., Kamibayashi, C., Robbins, D., Cobb, M., A88, 851-858. & Mumby, M. (1993) Cell 65, 887-897. Hu, S.-H., Parker, M. W., Lei, J. Y., Wilce, M. C. J., Benian, G. Taylor, S. S., Knighton, D. R., Zheng, J., Sowadski, J. M., Gibbs, M., & Kemp, B. E. (1994) Nature 369, 581-584. Hubbard, S. R., Wei, L., Ellis, L., & Hendrickson, W. A. (1994) C. S., & Zoller, M. J. (1993) Trends Biochem. Sci. 18, 84-89. Nature 372, 746-753. Witt, J. J., & Roskoski, R. (1980) Biochemistry 19, 143-148. Jones, T. A., You, J. Y., Cowan, S. W., & Kjelogaard, M. (1991) Xia, Y., Hwang, L.-Y., Cobb, M. H., & Baer, R. (1994) Oncogene Acta Crystallogr. A47, 110-115. 9, 1437-1446. Jove, R., Kornbluth, S., & Hanafusa, H. (1987) Cell 50, 937-943. Xu, S., Robbins, D., Frost, J., Dang, A., Lange-Carter, C., & Cobb, Kamps, M. P., & Sefton, B. M. (1986) Mol. Cell. Biol. 6, 751- M. H. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 6808-6812. 757. Zhang, F., Robbins, D. J., Cobb, M. H., & Goldsmith, E. J. (1993) Knighton, D. R., Zheng, J., Ten Eyck, L. F., Ashford, V. A., Xuong, J. Mol. Biol. 233, 550-552. N.-H., Taylor, S. S., & Sowadski, J. M. (1991) Science 253, Zhang, F., Strand, A., Robbins, D., Cobb, M. H., & Goldsmith, E. 407-414. J. (1994) Nature 367, 704-711. Lange-Carter, C. A., Pleiman, C. M., Gardner, A. M., Blumer, K. Zoller, M. J., & Taylor, S. S. (1979) J. Biol. Chem. 254, 8363- J., & Johnson, G. L. (1993) Science 260, 315-319. 8368. Lin, L.-L., Wartman, M., Lin, A. Y., Knopf, J. L., Seth, A., & Zoller, M. J., Nelson, N. C., & Taylor, S. S. (1981) J. Biol. Chem. Davis, R. J. (1993) Cell 72, 269-278. 256, 10837-10842. Mansour, S. J., Matten, W. T., Hermann, A. S., Candia, J. M., Rong, S., Fukasawa, K., Vande Woude, G. F., & Ahn, N. G. (1994) Science 265, 966-970. BI952723E

+ +